A molecular orbital analysis of electronic structure and bonding in

Variable-Energy Photoelectron Spectroscopy of CpM(CO)2 (M = Co, Rh, Ir): Molecular Orbital Assignments and an Evaluation of the Difference in Ground-S...
0 downloads 0 Views 732KB Size
7132 labeled complex exist. As was the case for ci-trans isomerization via hlWV,the identities of the a-a and e-e isomers merge while that of the a-e isomer remains distinct. The final conclusion, therefore, is that no single cis-trans isomerization mechanism can lead to interconversion of all three permutational isomers of a rigid cis-M(A-A’)*BZ complex, assuming of course that cis-trans isomerization does not involve any further symmetric intermediate configurations having connectivities greater than two. For the general case of generating a complete set of permutational isomers of a partially labeled compound having geometry V, one must proceed in the following manner: 1. Label the nuclear sites in all relevant configurations. 2. Label the nuclei which may occupy them. 3. Define and read into the computer the permutations in the proper configurational symmetry group VV,an improper configurational symmetry operation dVv (if the molecule has an achiral skeleton), the nuclear symmetry group L, and the group of allowed permutations Ifvv. 4. Read in the statement “ISOCNT V, H,L, VPRIME’. If the molecule has a chiral skeleton and dVv is ‘ undefined, the permutation dVvis omitted from the input. Supplementary Material Available: Complete description of the dynamic stereochemistry program and instructions for its use (23 pages). Ordering information is given on any current masthead page.

References and Notes (1) Parts 1 and 2 of this series: W. G. Klemperer, J. Am. Chem. SOC., 95, 380-396,2105-2120 (1973).

(a) University College London; (b) Columbia Unlversity; (c) Princeton University Computing Center. (a) W. G. Klemperer, J. Chem. Phys., 56, 5478-5489 (1972); (b) Inorg. Chem., 11, 2668-2678 (1972); (c) J. Am. Chem. SOC., 94,6940-6944 (1972); (d) ibid., 94,8360-8371; (e) W. G. Klemperer in “Dynamic Nuclear Magnetic Resonance Spectroscopy”, L. M. Jackman and F. A. Cotton, Ed., Academic Press, New York. N.Y., 1975, pp 23-44. (a) E. L. Muetterties, J. Am. Chem. SOC.. 91, 1636-1643 (1969); (b) P. (4) Meakin, E. L. Muetierties, F. N. Tebbe, and J. P. Jesson, ibid.. 93, 4701-4709 (1971). (a) E. Ruch, W. Hasselbarth,and B. Richter, 7bor. Chim. Acta, 19,288-300 (1970); (b)W. Hasselbarth and E. Ruch, ibid., 29, 259-267 (1973). C. K. Johnson and C. J. Collins, J. Am. Chem. SOC., 96, 2514-2523 (1974). (a) J. G. Nourse and K. Misiow, J. Am. Chem. Soc.,97,4571-4578 (1975); (b) J. G. Nourse, Proc. Natl. Acad. Sci. U.S.A., 72, 2385-2388 (1975). J. Brocas, Top. Cum. Chem., 32, 44-61 (1972). A complete description of the dynamic stereochemistry program DYNAMSTER employed below Is given in the Appendix together with instructions for its use. See paragraph at end of paper regarding supplementary material. T. J. Katz and C. R. Renner, unpublished results. I. Ugi, D. Marqwrding, H. Klusacek, G. Gokel, and P. Gillespie, Angew. Chem., Int. Ed. EngI., 9, 703-730 (1970). A convenient verbalization of an operation such as (143X2)6Ais “the nucleus in site 1A is moved to site 48, the nucleus in site 4A is moved to site 38, the nucleus in site 3A is moved to site 18, and the nucleus in site 2A is moved to site 28”. We shall refer to reactions representing different steric courses as differentiable reactions’ (or reactions differentiable in a chiral environment) and reactions representing the same steric course as nondifferentiable reactions’ (or reactions nondifferentiable in a chiral environment). When treating permutational isomerization reactions, other avthors have referred to a set of nondifferentiable reactions as a model4 or ring p e r m ~ t a t i 0 n . l ~ Within their limited range of application, these terms differ in no way from the more general and precise terminology employed here. J. I. Musher, J. Am. Chem. SOC.,94,5662-5665 (1972). J. A. Barltrop, A. C. Day, P. D. Moxon, and R. R. Ward, J. Chem. Soc.,Chem. Commun., 788-787 (1975), and references cited therein. W. G. Klemperer, J. K. Krieger, M. D. McCreary, E. L. Muetterties, D. D. Traficante, and G. M. Whitesides, J. Am. Chem. SOC.. 97, 7023-7030 (1975). (17) R. H.Holm, “Dynamic Nuclear Magnetic Resonance Spectroscopy”, L. M. Jackman and F. A. Cotton, Ed., Academic Press, New York, N.Y., 1975, pp 333-338.

A Molecular Orbital Analysis of Electronic Structure and Bonding in Chromium Hexacarbonyl Jeffrey B. Johnson and W. G. Klemperer* Contribution from the Department oJChemistry, Columbia University, New York. New York 10027. Received February 14, 1977

Abstract: The SCF-Xa-MSW method is used to calculate molecular orbitals in Cr(C0)6, the ligand array (CO)6, and free CO. Calculated ionization energies for Cr(C0)6 and CO are compared with experiment, as are calculated electronic transition energies for Cr(C0)6. Correlation of orbital energies, charge distributions, and wave function contours in these three systems leads to the conclusions that (1) metal-carbon bonding in Cr(C0)6 is due primarily to Cr3d-CO5u interactions, with only a minor contribution from C r 3 d - C O 2 ~interactions, (2) Cr4s and 4p orbitals have negligible bonding interactions with ligand orbitals, and (3) Cr3d-CO2a interactions, although weak relative to Cr3d-CO5u interactions, play a dominant role in determining the C - 0 bond order in Cr(C0)6 relative to free CO. Vibrational data, in particular interaction displacement coordinates, provide strong support for these conclusions.

An understanding of the interactions between carbon monoxide and transition metal atoms is essential to the understanding of structure and bonding in discrete organometallic carbonyl complexes and carbon monoxide adsorbed on metal surfaces. Chromium hexacarbonyl provides an ideal prototype system for the study of these interactions for several reasons: (1) Its octahedral symmetry implies equivalent orbital interactions at each ligand site. ( 2 ) Its symmetry also allows complete separation of metal-ligand d-a and d-a interactions. (3) The relative simplicity of Cr(C0)6 leads to photoelectron and ultraviolet spectra which are more readily interpreted than those of more complex systems. (4) A complete vibrational Journal of the American Chemical Society

99:22

analysis of Cr(C0)6 has been carried out which provides detailed information regarding bonding relationships in the molecule. Molecular orbital analyses of transition metal carbonyls have traditionally followed two general approaches. Following the first approach, molecular orbitals are generated using symmetry arguments and qualitative perturbation molecular orbital theory.’ This approach has gained wide acceptance owing to its simplicity and transferability. The principles involved are of sufficient simplicity and generality to be readily transferred from system to system. The major drawback of this approach, however, is its dependence on preconceived notions

1 October 26, 1977

7133 Table I. Calculated Eigenvalues, Charge Distributions, and Ionization Energies for Carbon Monoxide Molecular Orbitals Orbital

-e, eV

C

0

2a 5a

1.2 8.5 10.8 14.7

30 50 16 18

19 8 53 68

IT

4a

Charge distribution0 OUTS

Ionization energy, eV Calcd Obsd

INT

33 31

18 12 21 1

11

13

13.5 16.2 20.6

14.0 16.9 19.7

a Below C and 0 are the percentages of charge in 0.788 and 0.768 8, radii spheres about the C and 0 atoms, respectively; below OUTS is the percentage charge outside of a 1.342 8, radius sphere surrounding the molecule; and below I N T is the percent charge in the intersphere region, defined to be the charge not accounted for in the previous columns. Vertical ionization energy from ref 10.

Table 11. Calculated Eigenvalues, Charge Distributions, and Ionization Energies for (CO)6 and Cr(C0)6 Molecular Orbitals -6

Orbital

eV

8tl,

0.7

2t2,

0.8

7t1, 6alg 2t2, 5ege 6ti, It!, lt2, 5t1, ltls

(Co)6 Charge distributionb TvDeO C 0 OUTS I N T 32

9

28

32

2a

39

21

0

39

1.9 2.3 2.7

2a 2r

14 2 21

11 1 24

19 32 2

56 64 53

7.3 8.8 10.1 10.3 10.6 10.6

50 5a 1~ la la la

64 48 17 17 23 19

6 18 59 57 45 52

0 0 1 1 1 1

29 34 23 25 31 29

-e,

Orbital 2t1, loti, 6e, 2t2, 3t2, 9tl, 9a1, 2t2ge

5a1, 4e, 4ti, 4a1,

11.5 15.0 15.1 15.1

5u 4u 40 40

44 17 17 18

10 75 74 71

1

4 3 4

46 4 6 7

8t1, lti, lt2, 7tl, lt2, 5e, Sal, 4e, 6t1, 7a1,

eV

TvDeu

Cr

Cr(C0)6 Charge distributionC C 0 OUTS INT 20 5 4 23 16 17

1

52 24 16 40 9 21 3

d

59

50 la la la la 5a 5a 4a 4a 4a

5 0 0 2 2 29 10 0 0 1

0.5 1.3 1.6 1.7 2.0 2.6 2.7

2a p d 27r 2a 2a s

6.2 9.8 10.7 10.8 11.3 11.3 11.3 12.8 15.5 15.6 15.6

0 1 54 0 23 0

IE, ev Calcd ExDtld

1

2 30 5 0 7 8 28

26 40 22 37 46 53 66

6

13

0

22

8.6

8.4

46 18 19 27 21 45 43 17 16 18

23 58 55 39 47 6 13 74 74 68

0 1 1 1

26 23 25 32 30 19 33 5 6 9

12.1 12.9 13.1 13.5 13.4 13.6 15.0 17.8 17.9 17.9

13.4 14-16 14-16 14-16 14-16 14-16 14-16 17.8 17.8 17.8

1

0 1 4 3 4

Principal metal or ligand orbital contribution. Below C and 0 are the percentages of charge in 0.820 and 0.8 14 A radii spheres about the carbon and oxygen atoms, respectively; below OUTS is the percentage charge outside of a 3.894 A radius sphere surrounding the cluster; and below I N T is the percent charge in the intersphere region, defined to be the percentage charge not accounted for in the previous columns. Computed as in note b, where the chromium, carbon, oxygen, and outer sphere radii are 1.107,0.814,0.821,and 3.901 A, respectively. From ref 1 IC. e Highest occupied orbital.

regarding the relative strengths of various orbital interactions. A unique set of molecular orbitals can only be generated once the relative contributions of r vs. .n bonding, metal s, p, and d orbital participation, and metal-ligand vs. ligand-ligand interactions have been evaluated. Such a quantitative evaluation is beyond the scope of qualitative molecular orbital theory which therefore cannot uniquely generate molecular orbitals, but only present plausible bonding schemes based on various assumptions regarding the relative importance of different orbital interactions. A second approach to molecular orbital analysis in carbonyls is based on the quantitative generation of molecular orbitals using nonempirical quantum mechanical calculations.2 In contrast with the qualitative approach, the accuracy and hence uniqueness of these results can be assessed through a comparison of calculated electronic properties with spectroscopic data, and errors in the calculations can often be detected and corrected. Here, however, a sacrifice has been made regarding transferability. In order to transfer the insight gained into one particular molecule to a different system, one must first understand the component orbital interactions responsible for the energy and charge distribution of each molecular orbital. This has proved to be a formidable and largely unsuccessful task owing to the difficulty of isolating the large number of fundamental interactions possible in transition metal carbonyls. In this paper we combine the above-mentioned approaches, utilizing quantitative SCF-Xa-MSW calculations not only to

generate molecular orbitals, but also to isolate and assess the magnitudes of individual orbital interactions responsible for the overall molecular orbital structure. As a result, uniqueness can be assessed without sacrificing transferability. We are concerned here solely with the molecular orbitals in chromium hexacarbonyl, and will extend the results to related systems in future publications. Calculated results for CO, (CO)6, and Cr(C0)6 are presented, analyzed, and compared with appropriate experimental photoelectron and UV spectroscopic data. Discussion follows in three parts. First, the bonding scheme implied by the calculated results is compared with experimental data obtained from vibrational spectra. Next, the present results are compared with the results of previous theoretical studies of Cr(C0)6. A brief discussion relating the present results to studies of carbon monoxide absorbed on metal surfaces follows. Computational Procedures. All computations utilized Johnson and Slater’s SCF-Xa-MSW method3 with overlapping sphere^.^ Experimental bond distances were used for C O = 1.13 A)5 and Cr(C0)6 ( d c r - c = 1.92 A, d c o = 1.17 f)Fcalculations. The (CO)6 calculation utilized the carbon and oxygen coordinates found in Cr(C0)6. Atomic sphere radii (see Tables I and 11) were obtained by optimizing atomic number spheres4b with respect to the virial ratio, obtaining -2T/V = 1.0002, 1.0015, and 1.0010 for CO, (CO)6, and Cr(C0)6, respectively. In each case, the outer sphere was chosen to be tangent with the outermost atomic spheres. Schwartz’s CYHFexchange parameters’ were used in atomic

Johnson, Klemperer

/ Structure and Bonding in Chromium Hexacarbonyl

7134 LIGAND ORBITALS

SYMMETRY

METAL ORBITALS

Figure 2. Octahedral symmetry adapted combinations of ligand orbitals are shown on the left, and the metal atom orbitals they may interact with are shown on the right. Only one orbital or orbital combination is shown from each degenerate set.

W

4u Figure 1. Wave function contours for selected carbon monoxide molecular orbitals. Solid and broken lines indicate contours of opposite sign having absolute values of 0.3,0.2, and 0.1, T h e molecule is oriented with carbon on the left and oxygen on the right.

sphere regions and the average of the atomic values weighted according to total number of valence electrons for each type of atom was used in the outer sphere and intersphere regions. In all calculations, spherical harmonies through 1 = 2 were used in carbon and oxygen sphere regions, and functions through 1 = 4 were used in chromium and outer sphere regions. All S C F calculations were converged to better than 0.02 eV for each level with all cores relaxed. Calculated energy values have been converted from Rydberg units to electron volts (1 Rydberg = 13.6058 eV).*

Results In order to separate ligand-ligand interactions from ligand-metal interactions, molecular orbitals were calculated first for an isolated C O molecule, then for the (CO)6 ligand array found in Cr(C0)6, and finally for the complete Cr(C0)6 molecule. Calculations for a free Cr atom are not reported because chromium in Cr(C0)6 has a 3d6 configuration whereas the free Cr atom has a 3d54s1configuration, making comparison of atomic spectroscopic data with atomic “valence state” properties meaningless. Furthermore, the large number of electronic states resulting from a 3d6 configuration cannot be adequately represented using one-electron molecular orbital theory as employed here. The calculated eigenvalues and charge distributions for CO molecular orbitals (MO’s) in the -0.5 to -16 eV range are tabulated in Table I and wave function contours are shown in Journal of the American Chemical Society

Figure 1. When the octahedral (CO)6 array is formed, C O a and a orbitals combine to form the sets of symmetry adapted orbitals shown in the left column of Figure 2, where each set is ordered according to the number of new nodal surfaces generated. Calculated eigenvalues and charge distributions for (C0)6MO’s in the -0.5 to - 16 eV range are tabulated in Table 11, and (C0)6MO’s are correlated with C O MO’s in Figure 3, the splittings occurring in each case according to the number of new nodal surfaces generated. Note that the CO 50 and 2 a orbitals interact strongly upon formation of (CO)6, whereas the C O 4a and 1x orbitals interact only slightly. This difference in behavior follows directly from the localization of the 5a and 2x orbitals near carbon and the localization of the 4 a and l a orbitals near oxygen in C O (see Table I and Figure 1). In (CO)6, the carbon-carbon distances are shorter than the oxygen-oxygen distances, leading to greater interactions between orbitals localized near carbon. Of interest is the fact that the tl, combination of C O 5a orbitals may mix with ti, combinations of C O l a and C O 2x orbitals, a point which we shall return to below. Note also that the unoccupied (CO)6 6alg and 8t1, orbitals, having much electron density in the outer sphere region (see Table 11), arise from interactions of high-lying C O orbitals. The symmetry allowed orbital interactions between (CO)6 and Cr in Cr(C0)6 are shown in Figure 2. Calculated eigenvalues and charge distributions for Cr(C0)6 M O s in the -0.5 to -16 eV region are tabulated in Table 11, and theseorbitals are correlated with the (CO)6 orbitals in Figure 3. The 2tZg, 9alg, 6e,, and 10t1, levels correlate with the chromium 3da, 4s, 3da, and 4p orbitals, respectively. The small systematic lowering of eigenvalues in Cr(C0)6 relative to (C0)b evident in Figure 3 is most probably a computational artifact resulting from the difference between sphere sizes employed in the two calculation^.^ Selected wave function contours are shown in Figure 4. By viewing Figures 3 and 4 and consulting charge distributions tabulated in Table 11, the relative magnitudes of the

/ 99:22 / October 26, 1977

7135

601

8ol

10 0

I4 01

Figure 3. Molecular orbital correlation diagram indicating the parentage of Cr(C0)6 orbitals from (CO)6 and C O in the -0.5 to - 16.0 eV orbital eigenvalue range. All orbitals with t > -5 eV are unoccupied.

various metal-ligand interactions may be assessed. Strong a interaction between the C r d o orbitals and the CO 5a based (CO)6 5e, orbitals leads to the strongly bonding 5e, and strongly antibonding 6e, orbitals in Cr(C0)6. The strength of this interaction results in an approximately 4 eV lowering of the Cr(C0)6 5e, level relative to the (CO)6 5e, level. The a interaction between C r d a orbitals and the C 0 2 a derived (CO)6 2t2, orbitals is apparent from the energy of the unoccupied Cr(C0)6 3t2, metal-ligand antibonding orbitals relative to the unperturbed (CO)6 2t2, level. The shapes of the Cr(C0)6 t2, orbitals shown in Figure 4 reflect the character of the various metal-ligand a interactions. As implied by simple first order perturbation theory, the Cr(C0)6 1t2gorbital is primarily CO la in character, with some bonding contribution from the C r d a orbital; the 2t2, orbital is primarily a Crda orbital perturbed by bonding interaction with C 0 2 a orbitals and antibonding interaction with C o l a orbitals; and the 3t2g orbital is primarily C 0 2 a in character, with antibonding contribution from a Crda orbital. Comparison of the Cr(C0)6 5e, and 2t2g wave function contours emphasizes the greater magnitude of Cr-C a bonding relative to a bonding: there is far greater charge density in the Cr-C bonding region in the 5e, orbital. The essentially negligible u interaction between Cr4s and Cr4p orbitals with C05aderived (CO)6 5al, and 6t1, orbitals is reflected in the energetic proximity of the Cr(C0)6 8alg and 8t1, levels with their (CO)6 counterparts. Wave function contours for the Cr(C0)6 8al orbitals outline what is essentially a nodeless combination o f C 0 5 a orbitals, only slightly perturbed by bonding interaction with the Cr4s orbital. Wave function contours for the Cr(C0)6 tiu levels are more complex than those of the remaining levels owing to the a-a mixing which is allowed by symmetry: the 7tlu level is primarily C o l a in character with some bonding contribution from C 0 5 u orbitals; the 8t1, level is derived primarily from C 0 5 a orbitals, with bonding contributions from the Cr4p and C 0 2 a orbitals and antibonding contributions from CO1a orbitals; and the 9t1, level is primarily C 0 2 a in character, with antibonding contributions from the C 0 5 u orbitals. Note that the outermost radial nodes in the Cr4s and 4p orbitals are well removed from the C r core, giving the 8alg and 8t1, orbitals the superficial appearance of Cr-C antibonding orbitals.

Figure 4. Wave function contours for selected Cr(C0)G molecular orbitals. Solid and broken lines indicate contours of opposite sign having absolute values of 0.100, 0.075, and 0.050. The innermost contours around Cr in the 8a1, and 8t1, orbitals have been omitted.

The accuracy of the calculated CO and Cr(C0)6 occupied MO's may be assessed by comparison of calculated transition state ionization energies with experimental vertical ionization energies (see Tables I and 11). Experimental 21.1 eV He(1) photoelectron bands are in general well and may be interpreted unambiguously with the exception of the Cr(C0)6 13-16 eV region. Here, comparison of detailed band structure obtained by different workers reveals a lack of consistency. In each case, however, three general features are evident: a narrow band a t 13.4 eV and a more intense, broad, unsymmetric band a t about 14 eV which has a broad shoulder a t 14.5-15.5 eV. The broad 14 eV band arises from Cr(C0)6 levels having CO 1~ character, the 13.4 eV band arises from the 8tlu (primarily C 0 5 a ) level, and the 14.5-15.5 eV shoulder arises from the 5eg ( C 0 5 u Cr3d) and 8alg (primarily C 0 5 a ) levels. Note that the different orbital characters of the levels responsible for the three features are reflected in the different relative band intensities obtained in the 40.8 eV He(I1) spectrum.11bIn the He(I1) spectrum, the 14 eV band is far more intense than the 13.4 eV band and intensity in the 14.5-15.5 region equals that a t 14 eV. Although several workers have attempted detailed interpretation of similar intensity variations in terms of M O character,I2 we feel that such analysis is unwarranted in the absence of angularly resolved spectra' l e and calculated intensity data. Instead, we merely wish to point out that the relative intensity variations support the assignment of the 13-16 eV region to three features originating from levels having distinct M O character. Note that the photoelectron data are also consistent with an orbital ordering scheme where the 5e, level lies above the C O l a derived levels (see below). The overall ordering of unoccupied orbitals in Cr(C0)6 can be evaluated by comparison of experimental ultraviolet absorption energiesI3 with calculated spin restricted transition state energies as shown in Table 111. Most of our assignments, except for the two lowest energy absorptions, are in agreement with those of Beach and Gray.I3 The intense absorptions at 5.48 and 4.44 eV are assigned to the orbitally allowed 2t2, 2tzu and 2tzg 9tlu charge transfer transitions, respectively. The latter transition is weaker since the 9tl, has appreciable

+

-

-

Johnson, Klemperer / Structure and Bonding in Chromium Hexacarbonyl

7136 Table 111. Calculated and Observed Electronic Transition Energies in Cr(COl6

Species

Energy, eV Calcdb ObsdC

Transition"

Table IV. Bond Distances and Stretching Frequencies in Carbon Monoxide"

co co+

Confign

CO'

(Sa)' ( 5 0 ) ' . , . (7a)'

co

(50) ' ( 2 ~ )

'

State

dco, 8,

" ~ 0cm-I , ~

XIBf X2Z+

1 .I28

BIZ'

1.120

2143 2184 2082

A'A

1.235 1.206

1489 1715

a 3 ~

" Data from ref 19. available.

votl.

1.1 15

Data for the b3Z+ state are not

character with only a small ir contribution. Before presenting physical data which uniquely support this contention, we shall reevaluate the assumptions implicit in the traditional synergic model. First, the assumption that a donation strengthens the C O bond whereas ir bonding weakens the bond. The dependence of dco and vco on orbital configuration in free CO can C 0 5 u and Cr4p character. The weak 6.31 and 4.83 eV ab- be evaluated using data given in Table IV. When an electron sorptions are assigned to the orbitally forbidden 2tzg 2t1, is removed from a 5a orbital forming either CO+ X2Z+ or a and 2t2, -.+ 3t2, transitions, respectively. Two metal to metal CO Rydberg state B2Z+,a very slight shortening of dco (0.013 transitions calculated at 5.2 and 4.6 eV are obscured by the 2t2, and 0.008 A, respectively) occurs, accompanied by slight 2t2u transition. The two lowest energy absorptions are as- changes in uco ( t 4 1 and -61 cm-', respectively). When a 5a signed to vibration components of the 2t2, 9a1, metal to electron is placed into a 2ir orbital yielding the Alir and a 3a metal transition. A considerable body of experimental data,14 states, substantial increases in dco ( t 0 . 1 0 7 and t0.078 A, however, is consistent with assignment of these lowest energy respectively) occur, accompanied by equally substantial absorptions to the 2t2, 6e, transition. The present calculachanges in vco (-645 and -428 cm-I, respectively). In short, tions could very well be in error on this assignment since the the C 0 5 a orbital is for all practical purposes a nonbonding 9a1, level, having considerable charge in the outer sphere reorbital whereas the C02ir orbital is strongly antibonding, and gion, should make the 2t2, 9al, transition energy quite as a result, strong Cr-Cu bonding implies little regarding dco sensitive to environmental effects. Comparison of solution and and vco whereas even weak Cr-Cir bonding may have signifgas phase data, however, does not bear out this prediction. The icant impact upon dco and vco. origin of the apparent error can be easily explained. If the Next, we turn to the traditional assumptions regarding the Xa-MSW calculation has slightly overestimated the Crdasynergic nature to M-C bonding, where in order to preserve C 0 5 u interaction and thus produced a 5e, bonding orbital electroneutrality, it is assumed that u interactions must be which lies slightly too low in energy ( < I eV), the corresponding compensated by ir interactions having comparable magnitude. 6e, antibonding orbital will lie about 1 eV too high. Such a From the shape of the C 0 5 u orbital (see Figure l ) , it is clear readjustment of the 5e, and 6e, levels would not alter signifithat substantial electron density is displaced away from the cantly the relative strengths of metal-ligand u and ir bonding carbon end of the molecule. The same feature is observed in discussed above. This ambiguity could be resolved by carrying the C 0 2 a orbital to a far lesser extent. Thus when Cr-C out spin unrestricted transition state energy and intensity bonding occurs and charge is concentrated in the region becalculations, and also by more fully resolving the photoelectron tween Cr and C, a weak a interaction is able to charge-comspectra in the 13- 16 eV region. pensate for strong u interaction, since Crda to C 0 2 a bonding leads so far more charge transfer than C 0 5 u to Crdu bonding Discussion of an equivalent magnitude. This pattern of charge transfer Traditionally, the transition metal-carbon monoxide bond may be qualitatively verified by examining the charge distrihas been viewed in terms of a synergic model where ligand to butions in the Cr(C0)6 2t2, and 5e, orbitals given in Table 11. metal u donation from the CO5u orbital is balanced by metal About 20% of the charge in the 2t2, level resides on CO, reto ligand a back-donation into the C 0 2 a ~ r b i t a l . This ~ ~ . ~ ~flecting a back-donation, whereas comparison of the (CO)6 model is derived largely from the observation that the C-0 5e, orbital with the Cr(C0)6 5e, orbital shows about 20% loss bond distance dco increases and the C-0 vibrational freof charge from CO as a result of Q donation. Owing to the quency vC0 decreases upon complexation to a transition metal. higher degeneracy of the 2 t a~bonding level relative to the Although the relative magnitudes of u and a bonding have been 5e, level, there is if anything a net flow of charge from metal disputed, formulations of the synergic model are invariably to carbon in spite of the greater strength of the u interaction. based on two assumptions: (1) that u donation from the C 0 5 u One must not, however, give quantitative meaning to such orbital implies a shortening of dco and an increase in vC0 observations, since much charge redistribution occurs in the since the C 0 5 u orbital is antibonding, whereas back-donation intersphere region, and sphere sizes differ slightly in the (CO)6 into the antibonding C027r orbital has the opposite effect, and and Cr(C0)6 calculations. (2) that u bonding interactions must be about equal to a Vibrational displacement interaction coordinates for bonding interactions if approximate electroneutrality is to be Cr(C0)6 reported by Jones" provide the most convincing achieved. The traditional synergic bonding model is consistent physical evidence for the predominantly u character of the with many physical observations. It must be pointed out for Cr-C bond. The interaction coordinate G)i gives the relative future reference, however, that vibrational interaction conchanges in thejth bond length resulting from stretching of the s t a n t for ~ ~Cr(C0)6 ~ are claimed to be inconsistent with this ith bond, a positive value indicating a weakening of the j t h model. l 8 bond and a negative value indicating a strengthening. In the The present calculations as described above imply a bonding following discussion, bonds will be identified using this labeling model in Cr(C0)6 where M-C bonding is predominantly u in scheme: The predominantly metal or ligand character of the excited state orbital is indicated in parentheses. Symmetry allowed transitions are followed by asterisks. Spin restricted transition state energy. From ref 13, using the relation8 1 eV = 8065.5 cm-I. Oscillator strengths are given in parentheses.

-

-

+

-

-

Journal of the American Chemical Society / 99:22 / October 26, 1977

7137

Lt

0,-C,-Cr

.I ./

/I

'

c:

-C,-0,

I

YC7

OC

0,

When a Cr-C bond is stretched, the effect on the C-0 and remaining Cr-C bonds is defined by the interaction coordinates ( C o - O o ) ~ r - ~=, -0.0446, (Ct-Ot)cr-c, = +0.019, (CcOc)cr-c, = +0.0040, (Cr-Ct)Cr-Co = -0.228, and (CrCc)cr-co = +0.0041. The relative changes in C-0 bond lengths resulting from a stretching of a Cr-C bond reflect the fact that C - 0 bond strength is affected predominantly by Cr-C a,not u, bonding.'* The relative changes in Cr-C bond lengths, however, are almost entirely trans directed, a fact which is inconsistent with substantial 7r contribution to total Cr-C bond order. On the contrary, it demonstrates the predominantly u character of the Cr-C bond. Stretching of the Cr-C, bond in Cr(C0)6 leads to C4L'molecular symmetry where the d,~-~2 and dZ2 metal d u orbitals are orthogonal and no longer degenerate. Since trans Cr-C, u bonding involves only participation of the d,2 orbital, participation of the d,2-y2 orbital being symmetry forbidden, and cis Cr-Cc u bonding involves principally the d,2-y2 orbital, substantial perturbation of the Cr-C, bond and only slight perturbation of the Cr-C, bond is anticipated and observed when Cr-C, is weakened. W e turn now to comparison of the Cr(C0)6Xa-MSW calculation reported here with the HF-LCAO calculation reported by Hillier and Saunders2aand the HSF-DV calculation reported by Baerends and R o s . These ~ ~ two calculations yield molecular orbital eigenvalue orderings which are identical except for the position of the 8alg level, which lies below the C O la derived levels in the HF-LCAO calculation and lies in the same range as the C O la derived levels in the HFS-DV calculation. In contrast to the Xa-MSW calculation, the HF-LCAO and HFS-DV calculations yield a large (> 1 eV) separation between the 7a1, level and the remaining CO 4a derived levels. This result is clearly inconsistent with photoelectron data. The overall ordering of levels in these two calculations is otherwise identical with that shown in Figure 3, except for the location of the 5e, level which lies above the CO In derived levels in the HF-LCAO and HFS-DV calculations. As mentioned above, this location is consistent with photoelectron data. A final item of interest is the relevance of the present study to binding of CO on metal surfaces. Considerable experimental

and theoretical effort has been focused on the degree of interaction of the C 0 5 u orbital with transitional metal surfaces evidenced by the relative ionization energies of the perturbed C 0 5 u orbital and the relatively unperturbed CO la orbital.20 In Cr(C0)6, the energy of the Crda-CO5a 5e, bonding orbital relative to the free C 0 5 u level is the result of two opposing interactions: ligand-ligand antibonding interactions and metal-ligand bonding interactions (see Figure 3). In the absence of such ligand-ligand interactions on a metal surface, one would expect a more pronounced crossing of the C 0 5 u and CO 1 7r derived levels.

Acknowledgment. W e are deeply indebted to Professor K. H. Johnson for numerous helpful discussions. References and Notes (1) M. Elian and R. Hoffmann, lnorg. Chem., 14, 1058 (1975), and references cited therein.

(2) For representativecalculations see (a) i. H. Hillier and V. R. Saunders, Mol. Phys., 22, 1025 (1971);(b) E. J. Baerends and P. Ros, hid., 30, 1735 (1975); (c) J. Demuynck and A. Veillard, Theor. Chim. Acta, 28, 241 (1973); (d) I. H. Hillier and V. R. Saunders, Chem. Commun., 642 (1971); (e) H. B. Jansen and P. Ros. Theor. Chim. Acta. 34.85 (1974). (3) (a) K. H. Johnson, A&. Quantum Chem., 7, i43 (1973);'(b)Annu. Rev. Phys. Chem., 26, 39 (1975). (4) (a) N. Roesch, W. G. Klemperer, and K. H. Johnson, Chem. Phys. Lett., 23, 149 (1973); (b) J. G. Norman, Mol. Phys., 31, 1191 (1976). (5) 0. R. Gilliam, C. M. Johnson, and W. Gordy, Phys. Rev., 78, 140 (1950). (6) A. Whitaker and J. W. Jeffrey, Acta Crystallogr., 23, 977 (1967). (7) K. Schwarz, Phys. Rev. B, 5, 2466 (1972). (8) R. C. Weast, Ed., "Handbook of Chemistry and Physics", 56th ed, CRC Press, Cleveland, Ohio, 1975, p F-235. (9) In both calculations, - 2 T N could be more fully optimized by altering sphere radius ratios, thus changing the magnitudes of the eigenvalues without altering the spacing between eigenvalues significantly. (IO) For UV PES of CO see D. W. Turner and D. P. May, J. Chem. phys.. 45,471 ( 1966). (1 1) UV PES of Cr(C0)6 have been measured by (a) D. W. Turner, C. Baker, A.

(12) (13) (14) (15)

(16) (17) (18) (19) (20)

D. Baker, and C. R. Brundle, "Molecular Photoelectron Spectroscopy", Wiley-lnterscience, New York, N.Y., 1970, p 370; (b) B. R. Higginson, D. R. Lloyd, P. Burroughs. D. M. Gibson, and A. F. Orchard, J. Chem. Soc., Faraday Trans. 2, 69, 1659 (1973);(c) B. R. Higginson, D. R. Lloyd, J. A. Connor, and I. H. Hillier, ibid., 70, 1418 (1974); (d)W. C. Price, unpublished data; (e) E. W. Plummer. unpublished data. See, for example, W. C. Price, A. W. Potts, and D. G. Streets in "Electron Spectroscopy", D. A. Shirley, Ed., North-Holland PublishingCo., Amsterdam, 1972, p 187. N. A. Beach and H. B. Gray, J. Am. Chem. SOC.,90,5713 (1968). C. J. Ballhausen and H. B. Gray in "Coordination Chemistry", Voi. I, A. E. Martell, Ed., Van Nostrand-Reinhold, Princeton, N.J. 1971, pp 59-64. F. A. Cotton and G. Wilkinson, "Advanced Inorganic Chemistry", 3rd ed. Wiley-lnterscience, New York, N.Y., 1972, pp 683-701, and references cited therein; L. Malatesta and S. Cenini. Zerovalent Compounds of Metals", Academic Press, New York, N.Y., 1974, pp 1-60, and references cited therein. M. B. Hall and R. F. Fenske, lnorg. Chem., 11, 1619 (1972). L. H. Jones, J. Mol. Spectrosc., 36, 398 (1970). L. H. Jones and B. I. Swanson. Acc. Chem. Res., 9, 128 (1976). P. H. Krupenie, Natl. Stand. Ref. Data Ser., Natl. Bur. Stand., No. 5 (1966). D. R. Lloyd, C. M. Quinn, and N. V. Richardson, Solidstate Commun., 20, 409 (1976), and references cited therein.

Johnson, Klemperer

1 Structure and Bonding in Chromium Hexacarbonyl